The present invention relates to a method of evaluating the effects of external stimuli, such as pharmaceutical drugs and environmental influences like fragrances, temperature, noise, light, etc. on a subject""s brain, and more particularly to a method of evaluating the effects of administering such stimuli on a subject""s brain using imaging techniques with positron emission tomography (PET).
Positron emission tomography (PET) is a radiotracer based method for producing images that quantitatively represent some biochemical property of the body (or portions of the body). In relation to this work, use of PET is confined to metabolic imaging of the brain. Although other methods are often used, the aspect of PET that is relevant to this particular work involves 2-fluoro-deoxyglucose (FDG) as the tracer in studies of cerebral metabolism and oxygen-15 labeled water (0-15) as the tracer in studies of cerebral blood flow. In general, FDG is used to estimate the rate of metabolism of glucose in different parts of the brain (Sokoloff, 1985) and provides data that represent integrated metabolic activity over a 20-40 minute period. 0-15 studies determine rate of blood flow in different parts of the brain with an integration period of 40-60 seconds. Given the different temporal demands of the two kinds of tracers, metabolic studies with FDG reveal relatively long-lasting effects or conditions (such as pathologies), whereas 0-15 studies are more sensitive to rapid, transient activity (such as sensory processes or cognition).
The problem that we are interested in is to determine how certain external stimuli or treatments affect cerebral metabolism. The external stimuli or treatments are usually drugs, but other interventions would be faced with the same considerations. For example, environmental influences such as fragrances, temperature, noise, taste, vibration, light and similar stimuli clearly effect cerebral metabolism. The basic paradigm which we use to study external stimuli or treatments is conceptually very simple: 1) measure metabolism without any external stimuli or treatment; 2) apply the external stimuli or treatment; 3) measure metabolism again; and 4) determine whether the measurement at step 3 is statistically different from the measurement at step 1. In actuality, there are a number of experimental difficulties that must be dealt with before this paradigm can be applied.
First of all, it is important to realize that all images provided by PET reflect every influence on the brain at the time of a study. All perceptions, movements, thoughts, and moods, as well as vegetative functions, have correlates in brain metabolism and blood flow, and these factors, which are always present, may obscure effects due to external stimuli or treatment. Even more critically, these factors may change in unknown ways in response to the external stimuli or treatment and hence the extent of their influence on observed metabolism becomes unpredictable. It can, therefore, be difficult to determine which features of an image are due specifically to the experimental treatment and which are secondary, due to some other change that occurs because of the treatment. The objective of much of the present work has been to develop ways of processing PET images to more easily identify metabolic effects that are due to a specific external stimuli or treatment.
Some of our earliest work involved the recognition that the condition of subjects at the time of a study might vary from subject to subject or even within the same subject at different times (Levy et al., 1987). Variation in the testing condition thus could make it difficult to isolate differences introduced by an external stimuli or treatment.
Accordingly, we have developed an appropriate standard condition for testing subjects. This condition is the visual monitoring task (VMT). The VMT requires that subjects watch a screen on which is projected either a bright light or a dim light. The lights are easily distinguished from each other. One light flashes at a varying interval of 4 to 7 seconds. The two lights are equally probable. We ordinarily test subjects for 3 to 4 blocks (96 total trials each block, about 10 minutes per block) with a slight break between blocks. Subjects are instructed to press a button every time the dim light flashes and to ignore the bright flashes (a very subtle point: the natural tendency is to respond to the bright light which is more salient; by making the dim light the target a slight increase in difficulty is introduced). A computer measures reaction time (RT) to each button press (expressed as median RT per block) and whether the press was correct (a dim light), false alarm (a bright light), or missing (dim light flashed but subject did not press the button). In some situations, the VMT includes a feedback system so that subjects could see how fast their RT""s to target flashes were. This produces more consistent RT""s (lower variance). The VMT differs from other tasks that are occasionally used in PET studies (Buchsbaum et al., 1992; Hazlett et al., 1993) in that it is extremely simple and undemandingxe2x80x94subjects can do this task even if they are very young, very old, or slightly affected by a drug. At the same time, successful performance of the task precludes extraneous mental activity.
In early drug/PET work, a common procedure was to use a fairly large dose of a drug in order to produce the largest practical metabolic or blood flow xe2x80x9csignalxe2x80x9d. We, however, immediately recognized that this would create a problem. Some of the drugs that we were planning to study (e.g., ethanol, diazepam) would likely incapacitate subjects to the point where they would not be able to perform the VMT adequately. However, we were convinced that any dramatic change in behavior as a result of taking a drug would be impossible to interpret (as an extreme example, subjects who are sleeping after a drink of ethanol should not be compared to waking subjectsxe2x80x94there would undoubtedly be differences, but these would not be due to the drug but the condition of the subjects). Therefore, use of the VMT as part of our drug studies necessarily limits the dose of some drugs that can be studied. Thus, we chose to sacrifice a large but confounded signal in order to get a small but clean signal.
We also recognized that some subjects became very competitive while performing the VMT, visibly trying to get the lowest possible RT. We thus recognized that the demands of the VMT affected different subjects differently and perhaps would affect them differently under various drug conditions and/or other external stimuli treatments. Therefore, (and as now used in the OMEI process) we eliminated the RT feedback. Instead, we explicitly adopted an exclusion criterion: any condition on which RT is not stable (operationally defined as deviating by more than 10% from a reference condition) must be discarded. Similarly, any subject who does not perform with at least 95% accuracy (combined hits and correct rejections) must be omitted. Because subjects are confined to a relatively narrow range of behavior, we refer to this phase of the process as a xe2x80x9cbehavioral clamp.xe2x80x9d
The VMT provides fairly good control of subject""s overt behavior and even of their inner behavior (thinking). However, it does nothing to control mood, another variable that could be different under reference versus external stimuli (e.g. drug treatment) conditions, but as with sleep in the behavioral domain, it would be incorrect to attribute metabolic changes to a drug. In order to minimize the contribution of mood to metabolic changes that we would observe, we introduced into the PET experiments a standard test procedure. We administer the Profile of Mood Statesxe2x80x94POMS (McNair et al., 1971), a brief self-administered adjective check list that has been shown to be sensitive to drug effects (de Wit et al., 1985; de Wit et al., 1986) and other external stimuli. POMS scale scores are determined before and after the placebo and before and after the administration of the stimuli treatment. The difference between these scores indicates how much mood changed as a result of the administration of the stimuli or treatment, as opposed to changes due to fatigue, boredom, etc. We recognize that mood is difficult to control, but by measuring it we can incorporate significant mood changes into our interpretation of metabolic changes. Where practical, this often involves separating subjects who change in mood from those who do not (or who change in the opposite direction) and creating different images of metabolic change for each group.
Having deliberately chosen to deal with relatively small signals due to our external stimuli or treatment, we were next faced with the problem of detecting those signals. The standard method of dealing with metabolic images in PET studies (prior to OMEI) consists of drawing anatomical regions of interest (ROIs) on the slices that the scanner provides; this is done under both reference and treatment conditions. In its more recent form (Gut et al., 1995a), the ROIs are drawn on each subject""s PI then applied to the PET images that are spatially correlated (in three dimensions) with the MRI. In either form, this method is relatively insensitive to small changes in brain metabolism (Fox, 1991).
1. Even with the best positioning techniques, there will be slight differences in positioning of subjects on different occasions. In the case of repeated 0-15 scans we have even noticed significant changes in subject positions (up to 5 mm) within the same session; this problem, of course, is exacerbated when metabolic studies occur in different sessions on different days. This means that slices of the brain in one condition will not correspond exactly to ROIs from another condition.
2. The problem of different slices is even more serious when looking at different subjects since anatomical differences will prevent definition of identical ROIs.
3. Even the best drawing of ROIs cannot perfectly define all regions identically in all subjects. Not only will experimenter error and biases be present, but differences in anatomical features will cause some variation in definition of ROIs.
4. Any ROI must necessarily include relatively unresponsive subregions (e.g., white matter, portions near boundaries of ventricles or external surfaces that incorporate different partial volume effects).
5. True physiological effects will often not fill an entire ROI, no matter how small the ROIs may be.
6. Because the ROIs are defined independently of each other, physiological effects that cross ROI boundaries may fail to show up in any one ROI even though an effect may be relatively large.
The first five of these considerations serve to add xe2x80x9cnoisexe2x80x9d to the signal that we would be trying to detect; the sixth effectively reduces the size of a signal even further. Nevertheless, to our knowledge all PET metabolic studies to date, including our own (de Wit et al., 1988; de Wit et al., 1991), have used this basic approach. This approach, as we and others demonstrated, can work in the sense of demonstrating robust effects. When combined with the behavioral clamp procedure, this approach can certainly provide interpretable metabolic images. At worst, it would only require the studying of a sufficiently large number of subjects to determine effects of any external stimuli (e.g. drug) treatment (this, of course, can be practically impossible, given the cost of PET studies).
Coincidentally with these metabolic studies, however, others were conducting studies of cerebral blood flow with 0-15. In these studies it was early recognized that small signals were involved and therefore more sensitive analytic approaches were required. Such approaches were developed in several laboratories (Fox et al., 1988; Fox and Mintun, 1989). Underlying these more sensitive methods were two conceptual shifts from the standard procedure.
The first as a recognition that PET was providing true physiological data, not anatomical data (Fox, 1991). It stands to reason, therefore, that the physiological data itself would be more sensitive than a priori anatomical features. In this sense, these methods were xe2x80x9cdata drivenxe2x80x9d. The second shift was an effective realization that the slice-based data provided by PET scanners are estimates of true metabolic activity in the brain. Alternative ways of estimating metabolic activity can be validly employed. Briefly, the PET brain could be conceived of as a whole, relatively smooth volume rather than a set of discrete slices. Of course there are assumptions and limitations in this volumetric approach, but they are not necessarily worse than those of the slice-based approach. Most important, the volumetric approach allows for different kinds of data manipulation in experimental settings.
Specifically, estimates can be made for the metabolic value at every point in the brain and the brain can all be transformed into a standard three-dimensional space. Several different methods have been developed for estimating all points in the brain volume. Basically, they all involve interpolation from measured slice centers to every point in the vicinity of the slice center. At present, only linear interpolations have been employed but other methods are being researched (Lin et al., 1988; Lin et al., 1989). Likewise, the transformation problem has been solved several times (Evans et al., 1987; Fox et al., 1988; Evans et al., 1991). Investigators in our laboratory have been most successful in developing the procedure for spatially correlating PET and MRI or x-ray CT images (Pelizzari et al., 1989). While not essential, this step is one more way of reducing noise due to imprecision of the spacial transformations. Although we have preferred ways of treating our data, we recognize that there are a number of comparably good techniques. The critical point for the OMEI procedure is that volumetric handling of the data with transformation into standard space is an essential part of the procedure.
The present invention involves brain-behavior relationships and methods for evaluating and measuring them using imaging techniques with positron emission tomography (PET). In particular we are concerned with methods that measure quantitative changes in blood flow, metabolism and ligand localization and binding. More specifically we have been involved with elucidating the effects of external stimuli such as drugs and other psychoactive compounds with abuse potential as well as environmental influences like fragrances, temperature, vibration, taste, noise, light and other sensory-perceived influences, and cognitive challenges concerning attention and memory. We have effected a method which enables us to measure regional metabolic changes in the brain and associated mood changes as a result of administering such external stimuli, particularly from a single-dose drug challenge, in a controlled behavioral state.
The method presents a new perspective inasmuch as it reveals the end-pathway of the external stimuli""s effect by the metabolic process involved. Thus, it demonstrates those regional brain areas which effect the functional changes induced by the external stimuli, particularly by the action of a drug or psychoactive compound. Further, it can characterize the metabolic changes in quantitative terms as to whether the regional metabolic change is relatively increased or decreased. This xe2x80x9cend-effectxe2x80x9d measure is particularly important since we have shown it is quite distinct from the site of localization of the radiolabeled drug, its ligand-binding characteristics or the neurotransmitter systems involved.
A series of studies indicate this is an effective in vivo means of characterizing the effects of external stimuli, particularly drugs, and thus can provide a new and valuable approach to drug development. Specifically, we see an application in devising effective and efficient strategies in the clinical phases of development; not least from the ability to rapidly obtain a measure of effectiveness by direct comparison with already characterized and available compounds. We believe the method can be advantageously applied in all three clinical phases of drug developmentxe2x80x94safety, efficacy and dosage. It can also be a means to determine the effects of external stimuli, such as drug combinations and examine synergy or inhibition. Our special interest is in psychoactive compounds that effect mood and behavior and their associated neuropsychiatric disorders, but as noted above the method is applicable to other types of external stimuli. However, the approach is also applicable in drugs targeted to broader range of syndromes and disorders in the brain. Similarly, we recognize the potential in determining drug side-effects including CNS changes arising from non-brain targeted pharmaceuticals.
In terms of drug development, the efficiency of the method is an outstanding attribute. Significant measures may be obtained from as few as eight subjects and comparative results provided with a matter of weeks. It is amenable to many variations in the drug testing procedure including measures of acute and chronic effects and alterations in dosage, scheduling and delivery. It will provide an important measure of drug effect since for the first time it will be possible to relate dosage and drug plasma levels to a quantitative measure of regional changes in the central nervous system. Similarly, when these measures are related to parenchymal organ function of blood biochemistry, then toxicity versus therapeutic efficacy can be quantitatively appraised.
These considerations indicate that the method can have a significant impact upon both the cost and rate at which new drugs can be developed and brought to market. Archived compounds can be efficiently re-evaluated and, perhaps most importantly, it will increase the number and type of new compounds that can be applied in the therapy of diseases affecting the brain.
The method involves (1) measuring cerebral metabolism of a subject""s brain prior to and in the absence of any external treatment with a stimuli such as a psychoactive compound; (2) administering the external stimuli such as a psychoactive compound to the subject; (3) controlling behavioral influences on the subject""s brain by subjecting the subject to a behavioral clamp; (4) measuring cerebral metabolism of the subject""s brain after administering the external stimuli (e.g. psychoactive compound) and during the behavioral clamp; and (5) determining any differences between cerebral metabolism prior to and in the absence of administering the external stimuli (e.g. psychoactive compound) and cerebral metabolism after administering the external stimuli (e.g. psychoactive compound).